3 ms·

Correct me if I'm wrong but Clojure nudges avoiding mutable state while Haskell uses type theory to enforce it at your leisure at compilation time. Clojure see

Correct me if I'm wrong but Clojure nudges avoiding mutable state while Haskell uses type theory to enforce it at your leisure at compilation time. Clojure seems like a great language, but it has a completely different focus. I'd say Idris is a contender instead, which is not lazy and has theory proving. Will it become popular in the wild? Of course not, Python & JavaScript or something even easier is going to eat the world.

nimih 6y ago
I think Clojure puts a similar amount of effort to Haskell in order to discourage mutable state: the core library makes the easy solutions the ones that use immutable data structures and pure functions, and you have relatively ergonomic escape hatches (State/ST monad, atoms/volatiles/transients) available if you need them. Haskell's compile-time checking of purity and explicit effects is obviously absent from Clojure, but at a high level I think the two languages are mostly aligned in how easy they think it should be to use mutable state[0]. That being said, this is just picking nits and I agree with your larger point that Clojure and Haskell provide very different sets of trade-offs, so one really isn't a good substitute for the other[1]. [0] edit: Clojure just uses the more Clojure-y approach of eschewing compile-time checks in favor of assuming the user is able to responsibly use mutable state and indicate via some side-channel the contract a given function adheres to. Rich Hickey--if he cared about type systems in the first place--would likely argue that there's nothing inherent in Haskell which prevents programmers from writing truly labyrinthine stateful code by performing their computation within the appropriate monad; I have certainly done so myself, on occasion. [1] Unless your only metric is "ability to get functional programming weenies to argue/chatter in HN comments," in which case they pretty interchangeable IME.
- danielscrubs 6y ago
  I have nothing against Clojure. I like dynamic typing when doing data science, and I like heavily enforced types when someone else dumps code in my lap.

nimih 6y ago

I think Clojure puts a similar amount of effort to Haskell in order to discourage mutable state: the core library makes the easy solutions the ones that use immutable data structures and pure functions, and you have relatively ergonomic escape hatches (State/ST monad, atoms/volatiles/transients) available if you need them. Haskell's compile-time checking of purity and explicit effects is obviously absent from Clojure, but at a high level I think the two languages are mostly aligned in how easy they think it should be to use mutable state[0]. That being said, this is just picking nits and I agree with your larger point that Clojure and Haskell provide very different sets of trade-offs, so one really isn't a good substitute for the other[1]. [0] edit: Clojure just uses the more Clojure-y approach of eschewing compile-time checks in favor of assuming the user is able to responsibly use mutable state and indicate via some side-channel the contract a given function adheres to. Rich Hickey--if he cared about type systems in the first place--would likely argue that there's nothing inherent in Haskell which prevents programmers from writing truly labyrinthine stateful code by performing their computation within the appropriate monad; I have certainly done so myself, on occasion. [1] Unless your only metric is "ability to get functional programming weenies to argue/chatter in HN comments," in which case they pretty interchangeable IME.

danielscrubs 6y ago
I have nothing against Clojure. I like dynamic typing when doing data science, and I like heavily enforced types when someone else dumps code in my lap.